Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Decouple ObjectStore from Reqwest #7183

Draft
wants to merge 10 commits into
base: main
Choose a base branch
from
Draft

Decouple ObjectStore from Reqwest #7183

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
e1ac985
Create HttpClient
tustvold Feb 23, 2025
337dcdc
WIP
tustvold Feb 23, 2025
895ac28
Decouple from reqwest
tustvold Feb 23, 2025
de8f5db
Hook up HttpConnector
tustvold Feb 23, 2025
8c51a8a
WIP
tustvold Feb 23, 2025
d7e80e5
Update other stores
tustvold Feb 23, 2025
9f70bfd
Format
tustvold Feb 23, 2025
1d6de17
Remove out of date comment
tustvold Feb 23, 2025
2d543dd
RAT
tustvold Feb 23, 2025
e65c4be
Test fixes
tustvold Feb 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 9 additions & 8 deletions object_store/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,24 +45,28 @@ walkdir = { version = "2", optional = true }

# Cloud storage support
base64 = { version = "0.22", default-features = false, features = ["std"], optional = true }
form_urlencoded = { version = "1.2", optional = true }
http = { version = "1.2.0", optional = true }
http-body-util = { version = "0.1", optional = true }
httparse = { version = "1.8.0", default-features = false, features = ["std"], optional = true }
hyper = { version = "1.2", default-features = false, optional = true }
md-5 = { version = "0.10.6", default-features = false, optional = true }
quick-xml = { version = "0.37.0", features = ["serialize", "overlapped-lists"], optional = true }
serde = { version = "1.0", default-features = false, features = ["derive"], optional = true }
serde_json = { version = "1.0", default-features = false, optional = true }
rand = { version = "0.8", default-features = false, features = ["std", "std_rng"], optional = true }
reqwest = { version = "0.12", default-features = false, features = ["rustls-tls-native-roots", "http2"], optional = true }
ring = { version = "0.17", default-features = false, features = ["std"], optional = true }
rustls-pemfile = { version = "2.0", default-features = false, features = ["std"], optional = true }
serde = { version = "1.0", default-features = false, features = ["derive"], optional = true }
serde_json = { version = "1.0", default-features = false, features = ["std"], optional = true }
serde_urlencoded = { version = "0.7", optional = true }
tokio = { version = "1.29.0", features = ["sync", "macros", "rt", "time", "io-util"] }
md-5 = { version = "0.10.6", default-features = false, optional = true }
httparse = { version = "1.8.0", default-features = false, features = ["std"], optional = true }

[target.'cfg(target_family="unix")'.dev-dependencies]
nix = { version = "0.29.0", features = ["fs"] }

[features]
default = ["fs"]
cloud = ["serde", "serde_json", "quick-xml", "hyper", "reqwest", "reqwest/json", "reqwest/stream", "chrono/serde", "base64", "rand", "ring"]
cloud = ["serde", "serde_json", "quick-xml", "hyper", "reqwest", "reqwest/stream", "chrono/serde", "base64", "rand", "ring", "dep:http", "http-body-util", "form_urlencoded", "serde_urlencoded"]
azure = ["cloud", "httparse"]
fs = ["walkdir"]
gcp = ["cloud", "rustls-pemfile"]
Expand All @@ -72,16 +76,13 @@ tls-webpki-roots = ["reqwest?/rustls-tls-webpki-roots"]
integration = []

[dev-dependencies] # In alphabetical order
futures-test = "0.3"
hyper = { version = "1.2", features = ["server"] }
hyper-util = "0.1"
http-body-util = "0.1"
rand = "0.8"
tempfile = "3.1.0"
regex = "1.11.1"
# The "gzip" feature for reqwest is enabled for an integration test.
reqwest = { version = "0.12", features = ["gzip"] }
http = "1.1.0"

[[test]]
name = "get_range_file"
Expand Down
34 changes: 23 additions & 11 deletions object_store/src/aws/builder.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@ use crate::aws::{
AmazonS3, AwsCredential, AwsCredentialProvider, Checksum, S3ConditionalPut, S3CopyIfNotExists,
STORE,
};
use crate::client::TokenCredentialProvider;
use crate::client::{HttpConnector, ReqwestConnector, TokenCredentialProvider};
use crate::config::ConfigValue;
use crate::{ClientConfigKey, ClientOptions, Result, RetryConfig, StaticCredentialProvider};
use base64::prelude::BASE64_STANDARD;
Expand Down Expand Up @@ -171,6 +171,8 @@ pub struct AmazonS3Builder {
encryption_customer_key_base64: Option<String>,
/// When set to true, charge requester for bucket operations
request_payer: ConfigValue<bool>,
/// The [`HttpConnector`] to use to make [`HttpClient`]
http_connector: Option<Arc<dyn HttpConnector>>,
}

/// Configuration keys for [`AmazonS3Builder`]
Expand Down Expand Up @@ -882,13 +884,23 @@ impl AmazonS3Builder {
self
}

/// Overrides the [`HttpConnector`], by default uses [`ReqwestConnector`]
pub fn with_http_connector<C: HttpConnector>(mut self, connector: C) -> Self {
self.http_connector = Some(Arc::new(connector));
self
}

/// Create a [`AmazonS3`] instance from the provided values,
/// consuming `self`.
pub fn build(mut self) -> Result<AmazonS3> {
if let Some(url) = self.url.take() {
self.parse_url(&url)?;
}

let http = self
.http_connector
.unwrap_or_else(|| Arc::new(ReqwestConnector::default()));

let bucket = self.bucket_name.ok_or(Error::MissingBucketName)?;
let region = self.region.unwrap_or_else(|| "us-east-1".to_string());
let checksum = self.checksum_algorithm.map(|x| x.get()).transpose()?;
Expand Down Expand Up @@ -925,11 +937,7 @@ impl AmazonS3Builder {
let endpoint = format!("https://sts.{region}.amazonaws.com");

// Disallow non-HTTPs requests
let client = self
.client_options
.clone()
.with_allow_http(false)
.client()?;
let options = self.client_options.clone().with_allow_http(false);

let token = WebIdentityProvider {
token_path,
Expand All @@ -940,16 +948,19 @@ impl AmazonS3Builder {

Arc::new(TokenCredentialProvider::new(
token,
client,
http.connect(&options)?,
self.retry_config.clone(),
)) as _
} else if let Some(uri) = self.container_credentials_relative_uri {
info!("Using Task credential provider");

let options = self.client_options.clone().with_allow_http(true);

Arc::new(TaskCredentialProvider {
url: format!("http://169.254.170.2{uri}"),
retry: self.retry_config.clone(),
// The instance metadata endpoint is access over HTTP
client: self.client_options.clone().with_allow_http(true).client()?,
client: http.connect(&options)?,
cache: Default::default(),
}) as _
} else {
Expand All @@ -964,7 +975,7 @@ impl AmazonS3Builder {

Arc::new(TokenCredentialProvider::new(
token,
self.client_options.metadata_client()?,
http.connect(&self.client_options.metadata_options())?,
self.retry_config.clone(),
)) as _
};
Expand All @@ -986,7 +997,7 @@ impl AmazonS3Builder {
region: region.clone(),
credentials: Arc::clone(&credentials),
},
self.client_options.client()?,
http.connect(&self.client_options)?,
self.retry_config.clone(),
)
.with_min_ttl(Duration::from_secs(60)), // Credentials only valid for 5 minutes
Expand Down Expand Up @@ -1039,7 +1050,8 @@ impl AmazonS3Builder {
request_payer: self.request_payer.get()?,
};

let client = Arc::new(S3Client::new(config)?);
let http_client = http.connect(&config.client_options)?;
let client = Arc::new(S3Client::new(config, http_client));

Ok(AmazonS3 { client })
}
Expand Down
64 changes: 35 additions & 29 deletions object_store/src/aws/client.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ use crate::aws::{
AwsAuthorizer, AwsCredentialProvider, S3ConditionalPut, S3CopyIfNotExists, COPY_SOURCE_HEADER,
STORE, STRICT_PATH_ENCODE_SET, TAGS_HEADER,
};
use crate::client::builder::{HttpRequestBuilder, RequestBuilderError};
use crate::client::get::GetClient;
use crate::client::header::{get_etag, HeaderConfig};
use crate::client::header::{get_put_result, get_version};
Expand All @@ -31,7 +32,7 @@ use crate::client::s3::{
CompleteMultipartUpload, CompleteMultipartUploadResult, CopyPartResult,
InitiateMultipartUploadResult, ListResponse, PartMetadata,
};
use crate::client::GetOptionsExt;
use crate::client::{GetOptionsExt, HttpClient, HttpError, HttpResponse};
use crate::multipart::PartId;
use crate::path::DELIMITER;
use crate::{
Expand All @@ -42,17 +43,15 @@ use async_trait::async_trait;
use base64::prelude::BASE64_STANDARD;
use base64::Engine;
use bytes::{Buf, Bytes};
use hyper::header::{
use http::header::{
CACHE_CONTROL, CONTENT_DISPOSITION, CONTENT_ENCODING, CONTENT_LANGUAGE, CONTENT_LENGTH,
CONTENT_TYPE,
};
use hyper::http::HeaderName;
use hyper::{http, HeaderMap};
use http::{HeaderMap, HeaderName, Method};
use itertools::Itertools;
use md5::{Digest, Md5};
use percent_encoding::{utf8_percent_encode, PercentEncode};
use quick_xml::events::{self as xml_events};
use reqwest::{Client as ReqwestClient, Method, RequestBuilder, Response};
use ring::digest;
use ring::digest::Context;
use serde::{Deserialize, Serialize};
Expand All @@ -67,7 +66,9 @@ const ALGORITHM: &str = "x-amz-checksum-algorithm";
#[derive(Debug, thiserror::Error)]
pub(crate) enum Error {
#[error("Error performing DeleteObjects request: {}", source)]
DeleteObjectsRequest { source: crate::client::retry::Error },
DeleteObjectsRequest {
source: crate::client::retry::RetryError,
},

#[error(
"DeleteObjects request failed for key {}: {} (code: {})",
Expand All @@ -82,30 +83,32 @@ pub(crate) enum Error {
},

#[error("Error getting DeleteObjects response body: {}", source)]
DeleteObjectsResponse { source: reqwest::Error },
DeleteObjectsResponse { source: HttpError },

#[error("Got invalid DeleteObjects response: {}", source)]
InvalidDeleteObjectsResponse {
source: Box<dyn std::error::Error + Send + Sync + 'static>,
},

#[error("Error performing list request: {}", source)]
ListRequest { source: crate::client::retry::Error },
ListRequest {
source: crate::client::retry::RetryError,
},

#[error("Error getting list response body: {}", source)]
ListResponseBody { source: reqwest::Error },
ListResponseBody { source: HttpError },

#[error("Error getting create multipart response body: {}", source)]
CreateMultipartResponseBody { source: reqwest::Error },
CreateMultipartResponseBody { source: HttpError },

#[error("Error performing complete multipart request: {}: {}", path, source)]
CompleteMultipartRequest {
source: crate::client::retry::Error,
source: crate::client::retry::RetryError,
path: String,
},

#[error("Error getting complete multipart response body: {}", source)]
CompleteMultipartResponseBody { source: reqwest::Error },
CompleteMultipartResponseBody { source: HttpError },

#[error("Got invalid list response: {}", source)]
InvalidListResponse { source: quick_xml::de::DeError },
Expand Down Expand Up @@ -272,7 +275,7 @@ pub enum RequestError {

#[error("Retry")]
Retry {
source: crate::client::retry::Error,
source: crate::client::retry::RetryError,
path: String,
},
}
Expand All @@ -290,7 +293,7 @@ impl From<RequestError> for crate::Error {
pub(crate) struct Request<'a> {
path: &'a Path,
config: &'a S3Config,
builder: RequestBuilder,
builder: HttpRequestBuilder,
payload_sha256: Option<digest::Digest>,
payload: Option<PutPayload>,
use_session_creds: bool,
Expand All @@ -307,8 +310,8 @@ impl Request<'_> {

pub(crate) fn header<K>(self, k: K, v: &str) -> Self
where
HeaderName: TryFrom<K>,
<HeaderName as TryFrom<K>>::Error: Into<http::Error>,
K: TryInto<HeaderName>,
K::Error: Into<RequestBuilderError>,
{
let builder = self.builder.header(k, v);
Self { builder, ..self }
Expand Down Expand Up @@ -408,7 +411,7 @@ impl Request<'_> {
self
}

pub(crate) async fn send(self) -> Result<Response, RequestError> {
pub(crate) async fn send(self) -> Result<HttpResponse, RequestError> {
let credential = match self.use_session_creds {
true => self.config.get_session_credential().await?,
false => SessionCredential {
Expand Down Expand Up @@ -446,13 +449,12 @@ impl Request<'_> {
#[derive(Debug)]
pub(crate) struct S3Client {
pub config: S3Config,
pub client: ReqwestClient,
pub client: HttpClient,
}

impl S3Client {
pub(crate) fn new(config: S3Config) -> Result<Self> {
let client = config.client_options.client()?;
Ok(Self { config, client })
pub(crate) fn new(config: S3Config, client: HttpClient) -> Self {
Self { config, client }
}

pub(crate) fn request<'a>(&'a self, method: Method, path: &'a Path) -> Request<'a> {
Expand Down Expand Up @@ -544,6 +546,7 @@ impl S3Client {
.send_retry(&self.config.retry_config)
.await
.map_err(|source| Error::DeleteObjectsRequest { source })?
.into_body()
.bytes()
.await
.map_err(|source| Error::DeleteObjectsResponse { source })?;
Expand Down Expand Up @@ -641,6 +644,7 @@ impl S3Client {
.idempotent(true)
.send()
.await?
.into_body()
.bytes()
.await
.map_err(|source| Error::CreateMultipartResponseBody { source })?;
Expand Down Expand Up @@ -683,17 +687,17 @@ impl S3Client {
// If SSE-C is used, we must include the encryption headers in every upload request.
request = request.with_encryption_headers();
}
let response = request.send().await?;
let checksum_sha256 = response
.headers()
let (parts, body) = request.send().await?.into_parts();
let checksum_sha256 = parts
.headers
.get(SHA256_CHECKSUM)
.and_then(|v| v.to_str().ok())
.map(|v| v.to_string());

let e_tag = match is_copy {
false => get_etag(response.headers()).map_err(|source| Error::Metadata { source })?,
false => get_etag(&parts.headers).map_err(|source| Error::Metadata { source })?,
true => {
let response = response
let response = body
.bytes()
.await
.map_err(|source| Error::CreateMultipartResponseBody { source })?;
Expand Down Expand Up @@ -756,7 +760,7 @@ impl S3Client {

let request = self
.client
.request(Method::POST, url)
.post(url)
.query(&[("uploadId", upload_id)])
.body(body)
.with_aws_sigv4(credential.authorizer(), None);
Expand All @@ -781,6 +785,7 @@ impl S3Client {
.map_err(|source| Error::Metadata { source })?;

let data = response
.into_body()
.bytes()
.await
.map_err(|source| Error::CompleteMultipartResponseBody { source })?;
Expand All @@ -795,7 +800,7 @@ impl S3Client {
}

#[cfg(test)]
pub(crate) async fn get_object_tagging(&self, path: &Path) -> Result<Response> {
pub(crate) async fn get_object_tagging(&self, path: &Path) -> Result<HttpResponse> {
let credential = self.config.get_session_credential().await?;
let url = format!("{}?tagging", self.config.path_url(path));
let response = self
Expand All @@ -821,7 +826,7 @@ impl GetClient for S3Client {
};

/// Make an S3 GET request <https://docs.aws.amazon.com/AmazonS3/latest/API/API_GetObject.html>
async fn get_request(&self, path: &Path, options: GetOptions) -> Result<Response> {
async fn get_request(&self, path: &Path, options: GetOptions) -> Result<HttpResponse> {
let credential = self.config.get_session_credential().await?;
let url = self.config.path_url(path);
let method = match options.head {
Expand Down Expand Up @@ -895,6 +900,7 @@ impl ListClient for Arc<S3Client> {
.send_retry(&self.config.retry_config)
.await
.map_err(|source| Error::ListRequest { source })?
.into_body()
.bytes()
.await
.map_err(|source| Error::ListResponseBody { source })?;
Expand Down
Loading
Loading